DESCRIPTIVE STUDY OF INDONESIAN AND KOREAN TIME EXPRESSION WITH LGGs Prihantoro*, Sébastien Paumier ** *DICORA, Hankuk University of Foreign Studies, Korea/ ** IGM, University of Paris-Est, France [email protected], [email protected] Abstract This paper aims at considering some problems encountered when we translate time expressions between Korean and Indonesian. We consider in this paper the time expressions corresponding to the questions of type what time or when, rather than those related to the questions of type how long or how often. Especially the time expressions referring to calendar, day and clock time in Indonesian and Korean are considered in this study. Our discussion underlines four different aspects in translating the time expressions between these two languages: word order, numeral system, abbreviation, and time segmentation. The LGG(Local Grammar Graph) model will be used in the description of these aspects. The result of our study can be used as a linguistic resource not only for the learning system of Indonesian and Korean, but also for the machine translation system between these two languages. 1. Introduction Indonesian is the official language of Indonesia and was politically created in October 28th 1928, as an effort of unifying Indonesian people, who was colonialized by Netherlands. At that time, Indonesian communicated mostly in their first language only. Indonesian is the development of Malay used in Riau (Sneddon, 2004:14), in order to fulfill the plural characteristic of around 250 million Indonesian people. In other hand, Korean is the official language of Korea, both South and North. There are about 78 million Korean speakers. In the 15th century, a national writing system was commissioned by Sejong the Great. The system is currently called Hangul. Prior to the development of Hangul, Korean people had used Hanja (Chinese characters) to write for over a millennium (Sohn, 1999:122). Recently, the bilateral relation between Indonesia and Korea is getting more intense. Those two countries exchange many things such as, trading objects, technology, knowledge, culture, and of course language. For this reason, the need of understanding Korean and Indonesian language is also increasing, as well as the translation between these two languages. Korean and Indonesian are not genealogically related language. The gaps between these two languages pose challenges for translation, especially when this translation task is performed by machine. 1 The computer scientists and language experts are working to provide solution for this problem. However, before coming to that part, it is important to understand the different aspects between Indonesian and Korean Languages. That is the purpose of this paper. The paper is organized as follows. Section 1 is a brief introduction to both Korean and Indonesian languages. Section 2 shows the scope of problems that we present in this paper. In Section 3, we discuss four different aspects of time expressions between these two languages. This paper will be ended with a short summary and future perspective in Section 4. 2. Scope of the Problems We here are based on one of the previous studies related to time expressions in Korean language, which was done by J.-S. Nam(2009a, 2009b). In her papers, she introduces the LGGs(Local Grammar Graphs) to describe and recognize Korean time expressions. From the observation on Nam‟s studies, this paper is focused on the comparative studies of „date‟, „day‟, and „clock‟ time expressions between Indonesian and Korean. Our discussion will be focused on exact time expressions rather than duration expressions, which are classified as WHEN class in Nam(2009a:69, 2009b:164). These expressions will not answer to the questions like “How much time do I need to study programming?” or “How many days will you spend in Hawaii?”. Instead, they correspond to the questions like “When did you study Korean Linguistics?” or “What time will you go to Hawaii?”. 3. Different aspects between Indonesian and Korean Languages Translation is not only one-on-one word form conversion from source language to target language. Beyond that, translation conveys message from source language to target language. Winter(1961:98) believed, in order to achieve ideal translation, a translator must be able to handle the gaps, either linguistic or non-linguistic gaps, between source and target languages. Consider the following example, where Google translator failed to achieve a qualified translation. Illustration 1. Google Translation In the above example, source language is Indonesian, and target language is Korean. The Indonesian 2 phrase in the input box means “Monday, January 29th, 9 o’clock in the morning”. Korean speakers might be able to guess what the translated result signifies in the output box, but they cannot obtain an appropriate translation. It should be translated into “2009 nyon 1 wol 29 il wolyoil achim 9 si”. In this paper, we underline four problems in translating „calendar‟, „day‟ and „clock‟ expressions from Korean to Indonesian and vice-versa. These problems are “word order, numeral system, use of abbreviation, and time segmentation”. Our description is done as a DAG(Directed Acyclic Graph) we can build by using UNITEX GraphEditor1. 3.1 Word Order The order of Korean time expressions is usually like <Year, Month, Date, Day, Hour, Minute and Second>. We can see that the expressions are structured from the widest to the narrowest time entities. In Indonesian, there are no rigorous rules in ordering time expressions, but common order is like <Day, Date, Month, Year, Hour, Minute and Second>. The table 1 summarizes the order of time expressions in both languages. We here notice the gap of word orders between Indonesian and Korean languages in general. Table 1. Word Order in Time Expressions in Indonesian and Korean Korean Year Month Date Day Hour Minute Second Indonesian Day Date Month Year Hour Minute Second The following example shows the translation of the English phrase “Monday, January the 2nd 2009, nine forty five and twenty second” into both Korean and Indonesian. (1) Korean : 2009 nyon 1 wol 2 il Wolyoil 9 si 45bun 20 cho 2009-year 1-month 2-date Monday 9-hour 45-minute 20-second (2) Indonesian : Hari senin, tanggal 2 bulan Januari tahun 2009, jam 9 45 menit 20 detik Day-Monday date-2 month-January year-2009 hour-9 45 minute 20-second The following Graphs represent the order of time expressions in both languages: 1 These graphs built in Unitex are precisely Local Grammar Graphs(LGGs) which are suggested by French computational linguist Maurice Gross. The system UNITEX is a corpus-based processor, using an automata technology. (Paumier, 2003) 3 Graph 1. Korean and Indonesian Time Expression from Calendar to Clock Time Expression Korean Indonesian We observe that the order of Korean time expressions is structured from the widest range (Calendar) to the narrowest one (clock). Whereas in Indonesian, the order starts from day, calendar and ends by clock time. These graphs show that it is possible to put only one time slot as well as to combine them. When we discuss time expressions, there are mainly two points: numbers and time slot units (i.e. minute, hour, year etc). In Korean, all expressions indicating time slots are located after numeral expressions, which is not the case in Indonesian. Some of them can be located after numeral expressions and some of them can be before them. In Indonesian, Arabic numbers cannot be used to indicate Days and Months, while they are forbidden in the sequences of Days in Korean. This is a complete list in Table 2: Table 2. Order of Numeral-Classifier in Time Expressions in Indonesian and Korean Indonesian Korean Classifier Numeral Classifier Classifier Numeral Classifier Second 20 detik 20 cho Minute 45 menit 45 bun Hour Jam 09 09 si Day Hari (senin+*Num) (wol+*Num) yoil Date Tanggal 02 2 il Month Bulan (Januari+*Num) 1 wol Year Tahun 2009 2009 nyon In English clock time expressions, we recognize some units like half or quarter. Such expressions are also available in Indonesian and in Korean, but the order of these units are different. In Indonesian, there are some expressions like setengah(=half) and seperempat(=quarter), while in Korean, there is only ban(=half). Let us consider the following examples for Indonesian and Korean clock expressions. 1: 30 1: 15 (3) Korean : Han si ban Ø One hour half Ø (4) Indonesian: Jam setengah satu Jam satu seperempat Hour half one Hour one quarter 4 The following LGGs are representing some of these time expressions: Graph 2. Korean and Indonesian Clock Expression 3.2 Numeral System In Indonesian, there is only one numeral system. This makes the translation from Korean to Indonesian easier. Different from Indonesian, Korean has two numeral systems. One is native Korean numeral system(limited to 1~ 99), and the other is a system adopted from Chinese numeral system(Hobin, 2001: 144, Lee, 2004: 77). These two numeral systems are used in different contexts. Regarding time expressions, the challenging part is not when numerals are represented in Arabic numbers, but when they are in word forms. In written language, Arabic numbers could be used frequently, which do not make difference in both native Korean and Chinese numeral systems. However, in spoken language, even though Arabic numbers are used in the text, they have to be pronounced in a different way according to 5 their correspondence with native Korean or Chinese numeral system. The following table shows the correspondence of Arabic numbers with these 2 systems in a given context, i.e. the nature of time slot units: Table 3. Correspondence of Numerals with Time slot units in Korean Time slot units Numeral System Year Chinese Month Chinese Date Chinese Day Chinese Hour Korean Minute Chinese Second Chinese In the above table, hour is the only part requiring native Korean numerals.